bin/104039: [patch] cp's SIGINFO handler percent complete status wraps around when the bytes copied exceeds size_t

[Geoffrey Giesemann]

>Number:         104039
>Category:       bin
>Synopsis:       [patch] cp's SIGINFO handler percent complete status wraps around when the bytes copied exceeds size_t
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    freebsd-bugs
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          sw-bug
>Submitter-Id:   current-users
>Arrival-Date:   Fri Oct 06 05:00:38 GMT 2006
>Closed-Date:
>Last-Modified:
>Originator:     Geoffrey Giesemann
>Release:        6.2-PRERELEASE
>Organization:
>Environment:
FreeBSD geoff.cs.rmit.edu.au 6.2-PRERELEASE FreeBSD 6.2-PRERELEASE #10: Thu Oct  5 14:44:56 EST 2006     root at geoff.cs.rmit.edu.au:/usr/obj/usr/src/sys/GEOFF  i386
>Description:
cp keeps a running total of the bytes copied, so that when it gets a SIGINFO it can print a percentage copied.

The variable is of type size_t, so when bytes copied exceeds 4GB, it wraps to a negative value. Changing the type to off_t fixes this.
>How-To-Repeat:
dd if=/dev/zero of=/somewhere/with/lotsa/space/file bs=512 count=10M
cp /somewhere/with/lotsa/space/file /somewhere/with/lotsa/space/file2
<hammer ^T>
>Fix:
Change type of wtotal to off_t:

--- /usr/src/bin/cp/utils_old.c Fri Oct  6 14:05:34 2006
+++ /usr/src/bin/cp/utils.c     Fri Oct  6 14:25:19 2006
@@ -64,7 +64,7 @@
        int ch, checkch, from_fd = 0, rcount, rval, to_fd = 0;
        ssize_t wcount;
        size_t wresid;
-       size_t wtotal;
+       off_t wtotal;
        char *bufp;
 #ifdef VM_AND_BUFFER_CACHE_SYNCHRONIZED
        char *p;

>Release-Note:
>Audit-Trail:
>Unformatted: